Curves lab mode and colour balance

Alan

Currently, I'm looking at Adam's tutorials on how to enhance colour. Within the Curves process tutorial it is suggested that if you select the lab colour mode then you need to firstly click on 0.5, 0.5 and then create a contrast curve in either the a or b channels. This appears equivalent to method I'm familiar with in PS lab mode where you first click on 0,0 and then create a contrast curve in either the a or b channels. The reason you place the first point in a particular position is to prevent colour imbalance.

However, these two approaches don't seem equivalent. The PS approach keeps the colour balance but the PI method introduces a colour imbalance. You can see clearly see the effect if you go back to RGB mode and look at the histogram after you've performed the contrast curve - the peaks of the RGB channels become misaligned.

So, what am I doing wrong ?

Alan

BlockHead

Hi Alan,

I do not think you are doing anything wrong. I agree with your observation and I think that the algorithm for transposing RGB->Lab in PI (at least in this tool) may be different than in Photoshop. I really do not know enough to say more. 

By the way, this is an important sentiment. If I don't know... I am not going to even attempt to make something up. If I think I have insight into something, I will sometimes expand on it- and still be 100% wrong at times...but If I have no idea... that is pretty much the end.

This is one of the reasons it took me 4 years to even feel comfortable to present any information on PI. lol

If someone DOES know the answer (and bolster the purpose of this forum)... please respond!

-the Blockhead 

Alan

Thanks Adam.  

This does seem very odd, since I don't see how you could use the lab mode to effectively enhance colours in PI if it's always going to create a colour imbalance in the image.  

Alan

Alan

Hi Adam

Just to let you know that I asked the question on the Pixinsight forum and Juan recommended using the CIE l*c*h* colour space.  Apparently, this is the equivalent of CIE l*a*b* but instead of using a rectangular co-ordinate model, as in Lab, it uses a polar co-ordinate system. 

To use the CIE lch colour space for increasing saturation within Pixinsight you select the "c" button in the curves transformation process. 

I'm now trying to figure out the optimum shape of the c curve for processing images.....

Alan